Diketopiperazine Formation in Peptides: X-Pro Cyclization and Truncation

What a diketopiperazine is

A diketopiperazine (DKP, formally a 2,5-diketopiperazine) is the smallest cyclic peptide: two amino acids joined head to tail into a six-membered ring. The ring is rigid and uncharged, so it behaves nothing like the linear dipeptide it came from.

The reaction that matters on the bench is intramolecular aminolysis. The free α-amine at the N-terminus folds back and attacks the carbonyl of the second peptide bond. Residues 1 and 2 leave as the ring; residues 3 onward are released as a shorter peptide with a new free N-terminus. No water is consumed or produced, so the two product masses add up to the parent mass.

This is a different reaction from N-terminal cyclization to pyroglutamate, which uses the side chain of residue 1, keeps the chain intact and only shifts the mass by 17 or 18 Da. Both need a free, unblocked N-terminus, which is why an acetylated sequence such as MT-2 or a sequence that already starts with pGlu, such as ARA-290, is out of scope for either route.

Sequence: why X-Pro leads the list

Proline in position 2 is the classic risk pattern. Its ring restricts the backbone and raises the share of the cis configuration at the X-Pro bond, which is the geometry that puts the N-terminal amine next to the carbonyl it has to attack. Ion-mobility and mass spectrometry work on Phe-Pro-(Gly)n-Lys peptides followed trans-to-cis isomerization and then spontaneous cleavage at the Pro-Xaa bond, with transition-state free energies of 94.6 to 101.5 kJ/mol and chain-length-dependent rates.^[1]

The identity of residue 1 is not neutral either. In a series of X-Pro model compounds, changing the residue before proline (Gly, Ala, Val, Phe, β-cyclohexylalanine, Arg) significantly changed the rate of ring formation at pH 7.0.^[2] Glycine and small flexible residues generally make closure easier; bulky side chains slow it.

Context matters as well as sequence. Comparing a protein with model peptides built from its own N-terminal residues, ring formation was slower in the folded protein than in the free tripeptide, which the authors relate to a different cis-trans balance of the X-Pro bond.^[3] In recombinant human growth hormone, the same chemistry removed the N-terminal Phe-Pro and left a truncated variant identified by sequence analysis, tryptic mapping and mass spectrometry.^[4]

pH, buffer and solvent

The attacking group has to be an uncharged amine, so the rate tracks the ionization of the N-terminal amino group. In the Phe-Pro model study the pH-rate profile was described by a kinetic pKa of 6.1 and three microscopic rate constants, with the unprotonated form far more reactive; phosphate and glycine buffers showed general base catalysis, so buffer species and concentration change the number.^[2] The ring itself was stable between pH 3 and 8 and hydrolysed back to the linear dipeptide outside that window.

Solvent can matter more than pH. For H-Ala-Pro-NH₂ as the trifluoroacetate salt, the first-order half-life fell from 20 days in methanol to 3.6 minutes in acetonitrile containing 0.02 mol/L triethylammonium acetate — an 8,000-fold acceleration from an aprotic solvent plus an alkylammonium carboxylate.^[5] An H-Ala-Ala-NH₂ control went from an unmeasurably long time to 1.3 days in the same medium. Aprotic solvents and residual amine salts after cleavage are therefore a real risk for a vulnerable N-terminus. See buffers and pH selection for aqueous work.

Worked arithmetic: masses and half-lives

Mass first. Take the tripeptide H-Ala-Pro-Met-OH used as a model for a protein N-terminus.^[3] Residue masses are Ala 71.0371, Pro 97.0528 and Met 131.0405 Da, so the parent monoisotopic mass is 71.0371 + 97.0528 + 131.0405 + 18.0106 = 317.141 Da. Ring closure gives cyclo(Ala-Pro) at 71.0371 + 97.0528 = 168.0899 Da and free methionine at 131.0405 + 18.0106 = 149.0511 Da. Those two add back to 317.141 Da, which is the arithmetic check that the split is a rearrangement and not a hydrolysis.

The same rule generalises: for a peptide of mass M with residues R₁ and R₂ at positions 1 and 2, the truncated chain appears at M − (R₁ + R₂) and the ring at R₁ + R₂. A peptide starting Gly-Pro loses 154.074 Da — a gap no isotope pattern explains. The peptide mass calculator gives both numbers.

Kinetics next. First-order loss means the fraction remaining is 2^(−t/t½). At the methanol half-life of 20 days above, 1 day leaves 96.6% intact (3.4% converted) and 7 days leave 78.5% (21.5% converted). At the acetonitrile/triethylammonium acetate half-life of 3.6 minutes, 15 minutes leave 5.6%.^[5] Quote solvent, salt, temperature and pH alongside any half-life; the same dipeptide spans nearly four orders of magnitude between those media.

The synthesis side: a classic SPPS impurity

In solid-phase peptide synthesis the exposed moment is the dipeptide-resin after the second Fmoc removal, when a free amine sits two residues from the linker. Carboxyl-catalysed intramolecular aminolysis was described as a solid-phase side reaction in 1972, and it releases chains from the support before the third residue is coupled.^[6]

Modern process work shows the same pattern in long targets. A mechanistic study of Tirzepatide synthesis found ring formation mainly during Fmoc removal and post-coupling holds of proline-containing intermediates, including self-deprotection in DMF, DMSO, NMP and acetonitrile without added piperidine, giving double-residue deletion impurities. An oxyma additive and lower storage temperature reduced it, and replacing Fmoc with Bsmoc eliminated those by-products in that study.^[7]

Base choice at the deprotection step is a second lever: replacing 20% piperidine in DMF with a dilute 2% DBU / 5% piperazine mixture in NMP gave a sharp reduction in ring formation in a sequence-dependent case.^[8] Identify the missing species before changing the chemistry, because an apparent loss of chain is not always this reaction. See the SPPS overview and the related deletion sequence impurity class.

Detection, handling and a record checklist

Analytically the change is easy to see once you look for it. On RP-HPLC the truncated chain is a new peak, while the small neutral ring often elutes early; the model-reaction method followed loss of starting material and appearance of products at once.^[2] In mass spectrometry the signature is a peak at M − (R₁ + R₂) together with a low-mass ring ion, and the truncated chain can be confirmed by sequence-level fragmentation.^[4] Pair that with area-percent purity: a drifting purity value with a growing early-eluting shoulder is worth a mass check.

• Sequence and termini. Residues 1 and 2 written explicitly, with a note on whether the N-terminus is free, acetylated or pGlu.
• Conditions. Solvent (protic or aprotic), buffer species and concentration, pH, temperature and the time the material spent in solution.
• Expected masses. Parent mass, M − (R₁ + R₂) for the truncated chain and R₁ + R₂ for the ring, calculated before the run.
• Chromatography. Gradient and detection wavelength, retention times of parent, truncated chain and any early peak, and a freshly prepared reference analysed the same day.
• Synthesis history. For synthetic material: linker and resin, deprotection base and contact time, any holds at the dipeptide stage, and storage temperature of intermediates.

Practical handling follows from the kinetics: keep vulnerable sequences dry and cold, make solutions shortly before use, and avoid long holds in aprotic solvents with residual amine salts. This guide describes general laboratory chemistry and published literature only. It is not a statement about the content or behaviour of any particular material. Related routes are covered in peptide storage and degradation and pyroglutamate formation; terms are defined in the glossary; questions can be sent to support.